One prompt: "normal" input into Stable Diffusion XL for 11 minutes perhaps. Before beginning, I vowed to not cherrypick since they might all be boring. As it was, no selection, every image generated included, was easy. The first image set made me think: wtf.
Apparently normal is associated with error-free, balanced, symmetric, banal. These images are sometimes far from that. Objects sprout merge morph and contort, free from the constraints of a phenomoenologically familar world. This is the new normal: conceptual soup, ontological fluidity, corrosion blossoms.
Single-prompt series: "esoteric" html | vimeo & "voynich algorithm" html | vimeo & "profound" html | vimeo & "intrinsic" html | vimeo & "rabbit with robot eyes" html | vimeo & "intriguing parameters" html | vimeo
Earlier experiments: I-Drops (made with Dalle-2 in May 2023): #1 & #2 & splash #1 & #2 & #3 & #2 & #4 & #5 & #6